Repairing Numerical Equations in Analogically Blended Theories Using Reformation

Cheng-Hao Cai; Alan Bundy

Repairing Numerical Equations in Analogically Blended Theories Using Reformation

Cheng-Hao Cai, Alan Bundy

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

The ABC system supports analogical abduction algorithms for knowledge transfer, e.g., existing logical rules are adapted, by reformation, into new rules by substituting symbols for similar ones. Although such knowledge transfer method can easily expand knowledge sets, it is likely to produce inconsistent knowledge, e.g., equations that do not fit target data in the real world. To solve this problem, we extend the classic reformation algorithm to repair numerical equations that violate target data. Equation reformation is achieved by weakening equation parameters when unblocking failed SLD-resolution proofs. The feasibility of numerical equation reformation is demonstrated by the automated repair of a faulty electrostatic force equation that is analogically transferred from the gravity equation

Original language	English
Title of host publication	Proceedings of the Human-Like Computing Workshop 2022
Editors	Alan Bundy, Denis Mareschal
Publisher	CEUR Workshop Proceedings (CEUR-WS.org)
Pages	18-23
Number of pages	6
Volume	3227
Publication status	Published - 2 Oct 2022
Event	The 3rd International Workshop on Human-Like Computing 2022 - Windsor, United Kingdom Duration: 28 Sept 2022 → 30 Sept 2022 Conference number: 3 https://ijclr22.doc.ic.ac.uk/hlc2022.html/index.html

Publication series

Name	Human-Like Computing Workshop 2022
Publisher	CEUR Workshop Proceedings
Volume	3227
ISSN (Electronic)	1613-0073

Workshop

Workshop	The 3rd International Workshop on Human-Like Computing 2022
Abbreviated title	HLC 2022
Country/Territory	United Kingdom
City	Windsor
Period	28/09/22 → 30/09/22
Internet address	https://ijclr22.doc.ic.ac.uk/hlc2022.html/index.html

Keywords / Materials (for Non-textual outputs)

Reformation
Theory repair
Analogical reasoning
Knowledge transfer

Access to Document

Repairing Numerical_CAI_DOA15072022_VOR_CC-BYFinal published version, 214 KBLicence: Creative Commons: Attribution (CC-BY)

http://ceur-ws.org/Vol-3227/Cai-et-al.PP6.pdfLicence: Creative Commons: Attribution (CC-BY)

The 3rd International Workshop on Human-Like Computing 2022 (Event)
Bundy, A. (Editor) & Mareschal, D. (Editor)
28 Sept 2022 → 30 Sept 2022
Activity: Publication peer-review and editorial work types › Editorial activity

Cite this

@inproceedings{ceb8ff3a7c734dc4a9f6be3b2b6f7574,

title = "Repairing Numerical Equations in Analogically Blended Theories Using Reformation",

abstract = "The ABC system supports analogical abduction algorithms for knowledge transfer, e.g., existing logical rules are adapted, by reformation, into new rules by substituting symbols for similar ones. Although such knowledge transfer method can easily expand knowledge sets, it is likely to produce inconsistent knowledge, e.g., equations that do not fit target data in the real world. To solve this problem, we extend the classic reformation algorithm to repair numerical equations that violate target data. Equation reformation is achieved by weakening equation parameters when unblocking failed SLD-resolution proofs. The feasibility of numerical equation reformation is demonstrated by the automated repair of a faulty electrostatic force equation that is analogically transferred from the gravity equation",

keywords = "Reformation, Theory repair, Analogical reasoning, Knowledge transfer",

author = "Cheng-Hao Cai and Alan Bundy",

note = "Funding Information: For the purpose of open access, the author has applied a Creative Commons Attribution (CC BY) licence to any Author Accepted Manuscript version arising from this submission. Publisher Copyright: {\textcopyright} 2022 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0); The 3rd International Workshop on Human-Like Computing 2022, HLC 2022 ; Conference date: 28-09-2022 Through 30-09-2022",

year = "2022",

month = oct,

day = "2",

language = "English",

volume = "3227",

series = "Human-Like Computing Workshop 2022",

publisher = "CEUR Workshop Proceedings (CEUR-WS.org)",

pages = "18--23",

editor = "Alan Bundy and Denis Mareschal",

booktitle = "Proceedings of the Human-Like Computing Workshop 2022",

url = "https://ijclr22.doc.ic.ac.uk/hlc2022.html/index.html",

}

Cai, C-H & Bundy, A 2022, Repairing Numerical Equations in Analogically Blended Theories Using Reformation. in A Bundy & D Mareschal (eds), Proceedings of the Human-Like Computing Workshop 2022. vol. 3227, Human-Like Computing Workshop 2022, vol. 3227, CEUR Workshop Proceedings (CEUR-WS.org), pp. 18-23, The 3rd International Workshop on Human-Like Computing 2022, Windsor, United Kingdom, 28/09/22. <http://ceur-ws.org/Vol-3227/Cai-et-al.PP6.pdf>

Repairing Numerical Equations in Analogically Blended Theories Using Reformation. / Cai, Cheng-Hao; Bundy, Alan.
Proceedings of the Human-Like Computing Workshop 2022. ed. / Alan Bundy; Denis Mareschal. Vol. 3227 CEUR Workshop Proceedings (CEUR-WS.org), 2022. p. 18-23 (Human-Like Computing Workshop 2022; Vol. 3227).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - Repairing Numerical Equations in Analogically Blended Theories Using Reformation

AU - Cai, Cheng-Hao

AU - Bundy, Alan

N1 - Conference code: 3

PY - 2022/10/2

Y1 - 2022/10/2

N2 - The ABC system supports analogical abduction algorithms for knowledge transfer, e.g., existing logical rules are adapted, by reformation, into new rules by substituting symbols for similar ones. Although such knowledge transfer method can easily expand knowledge sets, it is likely to produce inconsistent knowledge, e.g., equations that do not fit target data in the real world. To solve this problem, we extend the classic reformation algorithm to repair numerical equations that violate target data. Equation reformation is achieved by weakening equation parameters when unblocking failed SLD-resolution proofs. The feasibility of numerical equation reformation is demonstrated by the automated repair of a faulty electrostatic force equation that is analogically transferred from the gravity equation

AB - The ABC system supports analogical abduction algorithms for knowledge transfer, e.g., existing logical rules are adapted, by reformation, into new rules by substituting symbols for similar ones. Although such knowledge transfer method can easily expand knowledge sets, it is likely to produce inconsistent knowledge, e.g., equations that do not fit target data in the real world. To solve this problem, we extend the classic reformation algorithm to repair numerical equations that violate target data. Equation reformation is achieved by weakening equation parameters when unblocking failed SLD-resolution proofs. The feasibility of numerical equation reformation is demonstrated by the automated repair of a faulty electrostatic force equation that is analogically transferred from the gravity equation

KW - Reformation

KW - Theory repair

KW - Analogical reasoning

KW - Knowledge transfer

M3 - Conference contribution

VL - 3227

T3 - Human-Like Computing Workshop 2022

SP - 18

EP - 23

BT - Proceedings of the Human-Like Computing Workshop 2022

A2 - Bundy, Alan

A2 - Mareschal, Denis

PB - CEUR Workshop Proceedings (CEUR-WS.org)

T2 - The 3rd International Workshop on Human-Like Computing 2022

Y2 - 28 September 2022 through 30 September 2022

ER -

Repairing Numerical Equations in Analogically Blended Theories Using Reformation

Abstract

Publication series

Workshop

Keywords / Materials (for Non-textual outputs)

Access to Document

Fingerprint

Activities

The 3rd International Workshop on Human-Like Computing 2022 (Event)

Cite this